behavioral/systems/cognitive ... · a cued articulation task con-sisting of 20 trials was the...

Behavioral/Systems/Cognitive

Predicting Conceptual Processing Capacity fromSpontaneous Neuronal Activity of the Left MiddleTemporal Gyrus

Tao Wei,1,2 Xia Liang,1 Yong He,1 Yufeng Zang,1 Zaizhu Han,1 Alfonso Caramazza,3,4 and Yanchao Bi1

1State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University, Beijing 100875, China, 2Department of Psychology, RiceUniversity, Houston, Texas 77251, 3Department of Psychology, Harvard University, Cambridge, Massachusetts 02138, and 4Center for Mind/Brain Sciences,University of Trento, I-38122 Trento, Italy

Conceptual processing is a crucial brain function for humans. Past research using neuropsychological and task-based functional brain-imaging paradigms indicates that widely distributed brain regions are related to conceptual processing. Here, we explore the potentialcontribution of intrinsic or spontaneous brain activity to conceptual processing by examining whether resting-state functional magneticresonance imaging (rs-fMRI) signals can account for individual differences in the conceptual processing efficiencies of healthy individ-uals. We acquired rs-fMRI and behavioral data on object conceptual processing tasks. We found that the regional amplitude of sponta-neous low-frequency fluctuations in the blood oxygen level-dependent signal in the left (posterior) middle temporal gyrus (LMTG) washighly correlated with participants’ semantic processing efficiency. Furthermore, the strength of the functional connectivity between theLMTG and a series of brain regions—the left inferior frontal gyrus, bilateral anterior temporal lobe, bilateral medial temporal lobe,posterior cingulate gyrus, and ventromedial and dorsomedial prefrontal cortices—also significantly predicted conceptual behavior. Theregional amplitude of low-frequency fluctuations and functionally relevant connectivity strengths of LMTG together accounted for 74%of individual variance in object conceptual performance. This semantic network, with the LMTG as its core component, largely overlapswith the regions reported in previous conceptual/semantic task-based fMRI studies. We conclude that the intrinsic or spontaneousactivity of the human brain reflects the processing efficiency of the semantic system.

IntroductionSemantic memory is a system for the storage, retention, and recall ofgeneral conceptual knowledge about objects, people, facts, and be-liefs that are unrelated to specific experiences (Tulving, 1972). Thissystem serves as a foundation for various cognitive processes includ-ing language, object recognition and use, reasoning, and problemsolving. Evidence from neuropsychological and task-based func-tional brain-imaging studies has shown that performing semantic/conceptual tasks (we do not intend to distinguish between semanticsand conceptual knowledge here, and use these two terms inter-changeably) implicates multiple areas in the temporal, frontal, andfrontoparietal regions (Dronkers et al., 2004; Binder et al., 2009).

To identify how conceptual knowledge is maintained at rest inthe absence of specific inputs or outputs, the current study ex-plores the role of spontaneous brain activity in semantic memory

using resting-state functional magnetic resonance imaging (rs-fMRI). We focus on object conceptual processing in relation tolow-frequency (�0.08 Hz) fluctuations (LFFs) in the blood oxy-gen level-dependent (BOLD) signal at rest, given that these fluc-tuations are related to spontaneous neuronal activity (Logothetiset al., 2001; Raichle, 2006).

Previous investigations of the regional amplitude of the LFFs(ALFF) have found that the ALFF reflects physiological signals:the ALFF of gray matter is higher than that of the white matter(Biswal et al., 1995), the ALFF of the so-called “default modenetwork” (regions that are activate during resting state and deac-tivated during task performance) is higher than that of otherregions (Zang et al., 2007), and individuals with cognitive braindisorders show abnormal ALFF in the regions critical for thecorresponding cognitive processes (Zang et al., 2007; Hoptman etal., 2010). Healthy individuals’ resting-state ALFFs correlate withtask-evoked BOLD responses and with participants’ behavioralmeasures (Mennes et al., 2011). Furthermore, high synchroniza-tion of LFFs between areas within the same neuroanatomicaland/or functional systems has been reported (Biswal et al., 1995;Fox et al., 2005; Koyama et al., 2010). Significantly, the degree ofsuch synchronization is associated with variability in healthy in-dividuals’ cognitive processing ability (e.g., reading ability)(Hampson et al., 2006), performance improvements (e.g., visual-detection performance) (Lewisa et al., 2009), and personalitytraits (e.g., autistic traits) (Di Martino et al., 2009). Thus, it ap-

Received April 19, 2011; revised Nov. 2, 2011; accepted Nov. 5, 2011.Author contributions: T.W., Y.H., and Y.B. designed research; T.W. performed research; T.W., X.L., Y.H., Y.Z., and

A.C. analyzed data; T.W., Z.H., A.C., and Y.B. wrote the paper.This work was supported by the National Natural Science Foundation of China (Grants 81030028, 30870667,

81020108022) and the Fundamental Research Funds for the Central Universities. A.C. was supported in part by theFondazione Cassa di Risparmio di Trento e Rovereto.

The authors declare no competing financial interests.Correspondence should be addressed to Yanchao Bi, State Key Laboratory of Cognitive Neuroscience and Learn-

ing, Beijing Normal University, Beijing 100875, China. E-mail: [email protected]:10.1523/JNEUROSCI.1953-11.2012

Copyright © 2012 the authors 0270-6474/12/320481-09$15.00/0

The Journal of Neuroscience, January 11, 2012 • 32(2):481– 489 • 481

pears that regional activity amplitude and connectivity patternsof the LFFs are associated with aspects of cognitive functioning.

In this study, we examined the extent to which intrinsic brainactivity (both regional activity and functional connectivity pat-terns) at rest predicts individual variation in semantic processingcapacity: (1) we performed a correlation analysis between partic-ipants’ performance on conceptual tasks and their regional ALFFto uncover potential core regions that could account for individ-ual variation in semantic processing efficiency; (2) we definedsuch core regions as seeds and computed the functional connec-tivity strength between each seed and other voxels to reconstructthe brain network associated with the core regions, and then weassessed the predictive power of these connectivity strengths forsubjects’ semantic performance. We found that regional ALFF inthe left (posterior) middle temporal gyrus (LMTG) and its func-tional connectivity with a set of other regions strongly predictedindividual variation in conceptual processing capacity.

Materials and MethodsParticipantsThirty-four healthy, right-handed college students (20 females; 22.5 �1.3 years old; range, 20 - 26 years old) with no history of neurological orpsychiatric disorders were recruited from the campus of Beijing NormalUniversity as paid participants, and written informed consents were ob-tained. This study was approved by the Institutional Review Board of theState Key Laboratory of Cognitive Neuroscience and Learning of BeijingNormal University.

Image acquisitionImages were acquired using a Siemens TRIO 3-Tesla scanner at the BeijingNormal University Imaging Center for Brain Research. The participants laysupine with their heads snugly fixed with straps and foam pads to minimizehead movement. Functional images were obtained using an echoplanar im-aging sequence with the following parameters: 33 axial slices; slice thickness,3 mm; gap, 0.6 mm; repetition time (TR), 2000 ms; echo time (TE), 30 ms;voxel size, 3.1 � 3.1 � 3.5 mm; flip angle, 90°; field of view (FOV),200 � 200 mm; and 240 volumes. In addition, a T1-weighted sagittalthree-dimensional magnetization-prepared rapid gradient echo se-quence was acquired: 128 slices; TR, 2530 ms; TE, 3.39 ms; slice thickness,1.33 mm; voxel size, 1.3 � 1.0 � 1.3 mm; flip angle, 7°; inversion time,1100 ms; FOV, 256 � 256 mm; and in-plane resolution, 256 � 256.During rs-fMRI scanning, participants were instructed to close their eyes,keep still, and not think about anything systematically or fall asleep.

Behavioral testsTo assess semantic processing ability, we focused on two commonlystudied object categories: animals and artifacts. We selected three classi-cal tasks that share semantic processing components but vary in themodalities of input and output: object picture naming, picture associa-tive matching, and object sound naming. By considering these tasks bothjointly and separately, we attempted to approximate the function of ashared semantic component and not just those of any specific input (e.g.,visual) or output (e.g., oral naming) process. We further administered aseries of other tasks to better establish that any potential results wereindeed relevant to the semantic component in those tasks. Two baselinecontrol tasks were used to regress out potential individual variation inperipheral motor and perceptual processing: a cued articulation task forthe naming tasks and a shape-matching task for picture associativematching. We also performed a number judgment task to test whetherthe regions/networks were specific for semantic processing or generalcognitive processing. The DMDX program (Forster and Forster, 2003)was used to present the stimuli and to record response latencies. Thebehavioral data were acquired about one year after image acquisition.

Object picture-naming task. Sixty color photographs of common ob-jects (30 tools and 30 animals) were used. Participants were instructed toname the pictures as quickly as possible without making errors. Each trialbegan with the appearance of a fixation point (“�”) on the center of thescreen for 500 ms, which was then replaced by a target picture. The

picture disappeared when the participant produced a vocal response orwhen a 3 s deadline was reached. The next trial started 1 s later. Picturesfrom different categories were randomized, with no consecutive trialsbeing semantically related.

Object picture associative-matching task. This task is a derivation of thePyramid and Palm Trees Test (Howard and Patterson, 1992). Each trialconsisted of a picture triplet with a reference picture (e.g., a hammer)displayed above a target (e.g., a nail) and a distracter picture (e.g., an axe).Participants needed to select the bottom picture that was more semanti-cally related to the top picture. Seventy-two picture triplets were used,including 36 tool triplets and 36 animal triplets. The trial structure was asfollows: A fixation point (“�”) was presented for 500 ms, followed by astimulus triplet. Participants responded by pressing a button box withtwo left-right aligned keys corresponding to the two alternative pictures.The triplets disappeared upon the response or after a 4 s deadline. Thenext trial started 1 s later.

Object sound-naming task. In this task, participants were required to namethe object that produces the target sound (e.g., “dog” for a barking sound).Typical sounds of 20 tools and 20 animals were selected. Each trial startedwith a one s fixation point (“�”), followed by a sound stimulus. The meanduration of the sound stimuli was 3.38 s (range, 0.73–12.02 s). The responsedeadline was 15 s. The next trial was initiated by the experimenter manuallyupon hearing the participants’ complete response.

Control tasks. Two nonsemantic control tasks that had similar taskstructures to the semantic tasks were used to regress out effects arisingfrom general response latency differences. A cued articulation task con-sisting of 20 trials was the control task for the naming tasks (i.e., objectpicture naming and sound naming). Participants were asked to pro-nounce the sound “ah” as soon as they saw the fixation point (“�”). Todiscourage adoption of response strategies, we used three randomizedtrial intervals: 500, 1000, or 2000 ms. A shape-matching task was thecontrol task for the picture associative-matching task. The procedure wasthe same as for the picture associative-matching task, except that geomet-ric shapes were presented, and the participants judged which of the twoshapes in the bottom was identical to the top target shape by pressing thecorresponding key. There were 36 trials.

Number judgment task. We administrated this task to assess the speci-ficity of the results obtained for the object conceptual tasks. Each trialcontained a reference number (e.g., 23) displayed above two other num-bers (e.g., 21 and 29), and participants were required to choose which ofthe latter two numbers was closer in magnitude to the reference number.There were 50 trials, and all items were two-digit numbers.

Data analysis proceduresBehavioral data analyses. To correct for speed-accuracy trade-off effects,we used an inverse efficiency (IE) measure—the average response time ofcorrect trials divided by accuracy—in the analyses of behavioral data(Townsend and Ashby, 1983). For normalization purposes, we com-puted the z scores of each participant’s IE; that is, we computed the meanIE for each participant and scaled each participant’s mean IE using the SDof the mean IEs across participants. Then we reversed the sign of the IE zscores so that higher scores would correspond to more efficient perfor-mance. This negated IE z score was used as the index for behavioralcapacity/efficiency throughout the analyses and is referred to as “effi-ciency score” below.

Functional imaging data preprocessing. Preprocessing was performed usingStatistical Parametric Mapping software (SPM8; http://www.fil.ion.ucl.ac.uk/spm) and Data Processing Assistant for Resting-State fMRI (Yan andZang, 2010). The first 10 volumes of the functional images were discarded forsignal equilibrium and adaptation of the participants to the scanning noise.Next, slice timing and head motion correction were performed, and a meanfunctional image was obtained for each participant. No participant exhibitedhead motion of �2 mm maximum translation or 2.5° rotation throughoutthe course of scans. To normalize functional images, each participant’s struc-tural brain image was coregistered to the mean functional image and wassubsequently segmented. The parameters obtained in segmentation wereused to normalize each participant’s functional image onto the MontrealNeurological Institute space (resampling voxel size was 3 � 3 � 3mm). After the linear trend of the time courses was removed, a band-

482 • J. Neurosci., January 11, 2012 • 32(2):481– 489 Wei et al. • Predicting Conceptual Behaviors from Resting-State Activity

pass filter (0.01– 0.08 Hz) was applied to reduce low-frequency drift andhigh-frequency noise. Finally, spatial smoothing (4 mm FWHM Gauss-ian kernel) was conducted to decrease spatial noise.

Because LFF is sensitive to signals in the gray matter, additional anal-yses were conducted on the gray matter mask generated using the follow-ing procedure. We included the voxels with a probability higher than 0.4in the SPM5 template onto the gray matter mask. Given the signal dis-tortion in cerebellum, we also excluded the cerebellar regions (#91–#116)in the Automated Anatomical Labeling template (Tzourio-Mazoyer etal., 2002). In total, there were 36,272 voxels in the gray matter mask.

Regional analysis: ALFF calculation and ALFF-behavior analysis proce-dures. Following Zang et al. (2007), the ALFF value of each voxel in thebrain was extracted as the sum of amplitudes within the low-frequencyrange (0.01– 0.08 Hz). Specifically, a filtered time series was transformedto the frequency domain to obtain a power spectrum for each voxel.Next, the square root was calculated at each frequency of the powerspectrum. The averaged square root across 0.01– 0.08 Hz at each voxelwas taken as the ALFF. We scaled the ALFF values for each participant forstandardization purposes: the mean ALFF value within the gray mattermask of the participant was computed, and then the ALFF of each voxelwas divided by this mean ALFF value.

We used two methods to approximate the semantic behavioral perfor-mance from the three semantic tasks. We examined whether regionalspontaneous activity was related to semantic processing capacity usingmeasures derived from both methods. The first approach (Method 1)was to obtain a composite score using the three semantic tasks and toconduct correlation analyses between this composite semantic score andthe ALFF (for a similar rationale, see Schwartz et al., 2009). The secondapproach (Method 2) was to consider the three semantic tasks separatelyand then to explore areas shared by these tasks.

In Method 1, we averaged the efficiency scores for the three semantictasks to generate a semantic composite score for each participant, whichwas taken as an index of the participant’s semantic processing capacity.We performed a regression analysis to reveal the relationship betweenALFF and semantic performance while partialling out the contributionof general response latency differences. Specifically, we regressed out theefficiency scores on the cued articulation and shape-matching controltasks from the semantic composite efficiency score and then computedthe correlation coefficient between the ALFF and the residualized seman-tic composite score. In Method 2, we explored any potential overlapamong the areas associated with different semantic tasks to identify thecommon region(s) for all three semantic tasks. To obtain a correlationmap for a specific semantic task, we calculated the correlations betweenthe ALFF and the residualized efficiency score in this specific task afterregressing out the efficiency scores in the corresponding control task. Forthe two naming tasks, the correlation map was obtained through a cor-relation between the ALFF and residualized efficiency scores in the nam-ing task after regressing out the efficiency scores in the cued articulationtask. Similarly, the correlation map for picture associative matching wasobtained through a correlation between the ALFF and residualized effi-ciency scores in the picture associative-matching task after regressing outthe efficiency scores in the shape-matching task. We then checked for thecommon regions for the three semantic tasks.

Network analysis: functional connectivity analysis and connectivity-behavioranalysis procedures. On the basis of the results of regional ALFF analyses, wefurther explored whether any observed core region works in concert withother regions as a network for semantic processing. Our rationale was to firstuse the observed region(s) as seed(s) to perform functional connectivityanalyses, mapping out the regions that function with the seeds as a network.Next, we examined whether or not any specific connections within this net-work were able to predict semantic behavior.

Before conducting functional connectivity analyses, six head motionparameters, white matter, and cerebrospinal signal were first regressedout. Functional connectivity analysis was performed using the Resting-State fMRI Data Analysis Toolkit (REST; http://www.restfmri.net). Aspherical seed ROI (radius, 6 mm) was created, centering on the coordi-nates of each peak point identified in the regional ALFF-behavior corre-lation analysis. To obtain the functional connectivity map for eachparticipant, we considered both of the two relevant measures: regression

coefficient (�) and correlation coefficient (r). First, we calculated themean time series of the seed ROI for each participant. For the same graymatter mask applied in the above regional analysis, we calculated theregression coefficients (�) and correlation coefficients (r) between theseed time series and other voxels to obtain a � map and an r map for eachparticipant. Fisher z score transformations were conducted for the corre-lation coefficients (r) to generate a z-functional connectivity (FC) map foreach participant. To identify the regions showing significant functional con-nectivity with the seed(s), we did one-sample t tests on these individual �maps or z-FC maps to see whether they were significantly different from zero(t � 6.43; corrected p � 0.01; Bonferroni correction). For these regions, weperformed correlation analyses between participants’ semantic performanceand � values or Fisher z scores while regressing out the peripheral responsecomponents reflected in the performance of the control tasks.

Validating the functional roles of the semantic network. To examinewhether the regions found in the ALFF- and FC-behavior analyses arespecific to (object) conceptual processing or are more generally involvedin other cognitive processes, we conducted the following ROI analyses.First, these regions and/or networks were defined as ROIs, and the cor-responding ALFF/FC values within each ROI were averaged. Then cor-relation analyses were performed between averaged ALFF/FC values andparticipants’ behavioral indexes in the number judgment task.

Reconstructing the default mode network. We compared the obtainedresting-state semantic network with the default mode network, which isactive during the resting state, is deactivated by most cognitive tasks but notby semantic tasks (Raichle et al., 2001), and overlaps in part with semanticnetworks derived from task-based fMRI research (Binder et al., 2009). Toidentify the default mode network in our participant group, we performedfunctional connectivity analyses (both correlation coefficient and regressioncoefficient measures) based on one previously identified seed region locatedin posterior cingulate gyrus (PCG; radius, 6 mm; central coordinates, �5,�49, 40) (Fox et al., 2005). Then, individuals’ correlation coefficients weretransformed into z-FC maps by the Fisher z score transformation. One-sample t tests were then performed on these individual � maps and z-FCmaps (t � 6.43; corrected p � 0.01; Bonferroni correction).

Statistical analysis. AlphaSim was used to correct for multiple compar-isons (originally in AFNI software and implemented in REST), and thecorrected threshold was set at p � 0.05. More specifically, the thresholdof regional ALFF-behavior analysis was the combination of a voxelwise pvalue of �0.05 and a cluster size of �58 voxels (1566 mm3). Becauseconnectivity-behavior analyses were conducted on brain areas showingsignificant functional connectivity with the seeds, the threshold was acombination of a voxelwise p value of �0.05 and a cluster size of �45voxels (1215 mm3) (small volume correction).

ResultsRegional ALFF analysis and ALFF-behavior analysisParticipants’ mean response times, accuracies, and inverse effi-ciency values on the behavioral tasks are shown in Table 1.

To explore the association between semantic behavior andregional resting-state activity, we first correlated the averagedefficiency scores of the three semantic tasks (Method 1) with theALFF value of each voxel across the whole brain. The ALFF of oneregion, the LMTG [peak, �60, �48, �3; 3564 mm 3 (132 voxels);Fig. 1A], was positively correlated with participants’ semanticperformance (rpeak � 0.68; rcluster � 0.81; p values �0.001; rpeak,correlation coefficient between the ALFF in the peak coordinateand behavioral scores; rcluster, correlation coefficient between theaveraged ALFF within the significant region and the behavioralscores; Fig. 1B). Thus, the regional ALFF variance of LMTG ac-counted for 66% (R 2) of the variation in semantic task perfor-mance across individuals.

When we conducted separate analyses on the three semantictasks (Method 2), the LMTG was the only brain region in whichthe ALFF was significantly positively correlated with (1) picture-naming performance with cued articulation performance re-gressed out [peak, �60, �48, �3; 2268 mm3 (84 voxels); rpeak �

Wei et al. • Predicting Conceptual Behaviors from Resting-State Activity J. Neurosci., January 11, 2012 • 32(2):481– 489 • 483

0.71; rcluster � 0.73; p values �0.001] and (2) picture associative-matching performance with shape-matching performanceregressed out [peak, �66, �45, �3; 3375 mm3 (125 voxels); rpeak �0.81; rcluster � 0.59; p values � 0.001]. In the object sound-naming task, we found positive correlations in right superiortemporal and inferior parietal lobes [peak, 54, �36, 24; 2511mm 3 (93 voxels); rpeak � 0.67; rcluster � 0.67; p values �0.001]. Itshould be noted that the particularly low levels of accuracy andlarge variance in performance in this task may have originated inpart at the object sound perceptual/recognition phase. Since wedid not include a control task for this phase of the object sound-naming task, we lowered the threshold to corrected p � 0.1 [acombination of a voxelwise p value of �0.05 and a cluster size of�1377 mm 3 (51 voxels)]. We found that,with cued articulation regressed out,LMTG showed association with objectsound-naming performance [peak, �66,�42, �6; 1431 mm 3 (53 voxels); rpeak �0.59; rcluster � 0.64; p values �0.001].

Together, the findings obtained withthe two methods converge in showing thatthe LMTG is relevant for semantic pro-cessing regardless of stimulus input andoutput modality.

Functional connectivity analysis andconnectivity-behavior analysisTo explore whether LMTG functions inconcert with other brain regions for se-mantic processing, we performed a seedvoxel correlation analysis between theLMTG region found with the semanticcomposite score (Method 1) and all othervoxels in the brain. As seen in Figure 2, Aand B, we found similar functional connectivity patterns usingthe regression coefficient and correlation coefficient measures. Awide range of brain regions showed significant functional con-nectivity with LMTG, including the bilateral temporal gyri, infe-rior occipital gyri, middle and inferior frontal gyri, precentralcortices, ventromedial and dorsomedial frontal cortices, angulargyri, and posterior cingulate gyri. We defined these regions as alarge ROI mask and correlated each voxel value within this ROIwith semantic composite scores across participants. For the re-gression coefficient approach, the strength of functional con-nectivity between LMTG and the following regionssignificantly predicted participants’ semantic performance:left inferior frontal gyrus (LIFG), bilateral lateral temporallobe including anterior temporal lobe (ATL), left medial tem-poral lobe (LMTL), PCG, dorsomedial prefrontal cortex(DMPFC), and ventromedial prefrontal cortex (VMPFC)(Fig. 3; Table 2). When the correlation coefficients were con-sidered, significant correlations were also observed betweenfunctional connectivity in most of these regions and LMTGand participants’ semantic performance (Table 2).

We performed ROI analyses to further examine whether theregional activity of these regions (in the regression coefficientresults) correlated with semantic processing performance. First,the averaged ALFF value within each region was obtained. Thencorrelation analyses between participants’ semantic compositescores and averaged ALFF values for each region were per-formed. No region showed significant effects in these ROIanalyses (r values �0.27; p values �0.12).

Combining regional- and connectivity-behavior analysesTo assess the joint contributions of the regional activity and con-nectivity of LMTG in predicting semantic behavior, we per-formed a multiple linear regression analysis. The dependentvariable was participants’ semantic composite scores after re-gressing out the efficiency scores on the cued articulation andshape-matching control tasks. Independent variables includedthe ALFF values of LMTG and seven functional connectivity vari-ables. The ALFF variable was obtained by averaging ALFF valueswithin the LMTG region that was significant in the regional anal-ysis (Method 1). The functional connectivity variables were ex-tracted by averaging the regression coefficients within eachobserved region in the functional connectivity-behavior analyses.When all independent variables were simultaneously enteredinto the regression, the regression model explained 74% of thevariation in participants’ semantic processing performance (R2 �0.74; F(8,25) � 8.96; p � 0.001). We also used a forward methodto test how much independent contribution each of the twotypes of variables provided. When the ALFF variable was en-tered first and the 10 functional connectivity variables wereentered in the second step, we found that the inclusion of thefunctional connectivity variables yielded a trend of improve-ment for the explanatory power of the regression model (R 2 �0.663 0.74), but the effect was not significant (F(7,25) � 1.18;p � 0.35). When the functional connectivity variables wereentered first, the addition of ALFF into the model yieldedsignificant R 2 change (R 2 � 0.493 0.74; F(1,25) � 24.18; p �0.001). These results further highlight the central role ofLMTG’s regional activity in predicting semantic performance.

Figure 1. Regional ALFF-behavior correlation analyses using composite semantic scores. A, Statistical map for the correlationbetween ALFF and semantic scores. The correlation value is indicated using the color scale to the top. B, Scatter plot shows thepositive correlation between semantic performance and averaged ALFF in the region plotted in A, while the efficiency scores on thecued articulation and shape-matching control tasks were regressed out from the semantic composite efficiency score. Each dotrepresents data from one participant.

Table 1. Participants’ performance on behavioral tasks

Tasks

Response time(ms) Accuracy

Inverseefficiency

Mean SD Mean SD Mean SD

Semantic tasksObject picture-naming 1082 152 94% 4% 1149 187Object picture associative-

matching 1608 243 91% 6% 1781 292Object sound-naming 2857 506 76% 10% 3832 968

Other tasksCued articulation 376 91 100% 0% 376 91Shape matching 487 72 96% 4% 506 64Number judgment 1316 310 93% 4% 1409 317


The functional specificity of the observed semantic networkThe results of ROI analyses for the number judgment task areshown in Table 3.a Within the LMTG network we obtained forthe semantic tasks, we observed that the intrinsic brain activity(i.e., both regional activity and functional connectivity) was notsignificantly associated with the number-processing task. These

results indicate that different domains of conceptual processing(object vs numbers) can be distinguished.

Default mode network and semantic networkAs shown in Figure 4A, bilateral inferior parietal lobe, medialprefrontal cortex, superior frontal cortex, medial temporal lobe,and lateral temporal cortex had strong functional connectivitywith PCG, and these regions constitute the default mode net-work. Figure 4B replots the data presented in Figures 3 and 4A,showing that the semantic network largely overlaps with the de-fault mode network.

aWe also performed whole-brain correlation analyses between the regional activity and efficiency scores in thenumber judgment task. We observed a significant association between participants’ behavior and intrinsic activityin left inferior and superior parietal lobule encompassing the left intraparietal sulcus [peak coordinates (range),�51 (�52��17), �48 (�59��38), 54 (47�67); 57 voxels; corrected p � 0.05], close to the numberregions reported in previous studies (Dehaene et al., 2003).

Figure 2. A, B, Statistical maps of functional connectivity of LMTG with the regression coefficient (A) and correlation coefficient (B) measures: voxels for which the time series showed a significantassociation with the seed ROI in LMTG.

Figure 3. Clusters for which functional connectivity strength (�) with LMTG significantly predicted semantic composite scores. Scatter plots show positive correlations between participants’semantic performance and functional connectivity strength between LMTG and the clusters in the blue circles, while the efficiency scores on the cued articulation and shape-matching control taskswere regressed out from the semantic composite efficiency score. Each dot represents data from one participant.


DiscussionWe observed that the amplitude of resting-state LFF activity inLMTG accounted for 66% of individual variance in object con-ceptual processing efficiency. This region functions together withsix other regions in a semantic memory network: we found thatthe strength of functional connectivity between LMTG and theseregions (LIFG, bilateral ATL, PCG, bilateral MTL, VMPFC, andDMPFC) also significantly predicted participants’ semantic pro-cessing efficiency. Regression analysis revealed that this network,with LMTG as its core component, accounted for 74% of seman-tic processing performance variation, but did not correlate withparticipants’ performance on a non-object-conceptual task (i.e.,number processing).

The LMTG network identified here largely overlaps with re-gions that have been previously implicated in semantic process-ing tasks in the functional brain-imaging literature. LMTG haslong been observed to be important for semantic processing, andall seven regions obtained in the network analysis overlap withthe regions that were reported in a previous meta-analysis oftask-based fMRI and positron emission tomography studies ofsemantic processing (Binder et al., 2009). The only difference isthat the meta-analysis reported a significant role for the angulargyrus, while this area was not found in our analysis.

LMTG as a core semantic regionThe LMTG region identified here corresponds to the middle partof Brodmann’s area (BA) 21, extending to BA 37. In brain-imaging studies, the LMTG has been found to be activated bytasks that probe semantic processing through various modalities,including visual words and pictures (Vandenberghe et al., 1996),auditory words (Hickok and Poeppel, 2004), and tactile inputs(Stoeckel et al., 2003). Task-induced activity in the LMTG has

also been observed for semantic information acquisition relativeto baseline (Maguire and Frith, 2004). The essential role of theLMTG in word-level comprehension is also indicated by lesion-symptom mapping analysis on stroke patients (Bates et al., 2003;Dronkers et al., 2004; Schwartz et al., 2009) and voxel-based mor-phometric analyses on semantic dementia patients (Mum-mery et al., 2000). Turken and Dronkers (2011) showedpreviously that the LMTG has rich structural and spontaneousfunctional connectivity with other regions relevant for lan-guage comprehension, and they proposed that it plays a cen-tral role in comprehension.

In some task-based fMRI and lesion studies (Damasio et al.,1996; Martin et al., 1996), the LMTG has been shown to be moresensitive to certain semantic categories such as tools, while otherstudies have observed effects for a wide range of semantic catego-ries (Rudrauf et al., 2008; Simmons et al., 2010), including verbs(Willms et al., 2011), suggesting that the effect reported here isnot likely to be driven only by tool items. We nevertheless ana-lyzed tool and animal items separately in our study, and foundsimilar LMTG effects for the two categories [for tools, peak, �63,�45, �3; 3024 mm 3 (112 voxels); rpeak � 0.64; rcluster � 0.77; pvalues �0.001; for animals, peak, �60, �48, �3; 3186 mm 3 (118voxels); rpeak � 0.63; rcluster � 0.74; p values �0.001]. It is possiblethat the tool-specific effects reported in the literature originatedfrom an LMTG region that was close to but different from ours(Simmons et al., 2010). We did show, however, that numberjudgment performance did not correlate with the observed objectsemantic LMTG network, but with parietal regions close to theclassical number regions (Footnote a), suggesting that ourtechnique has the potential to distinguish between conceptualdomains.

Together, these findings constitute clear evidence for a criticalrole of the LMTG in semantic processing, at least for single objectconcepts (Lau et al., 2008). The present study further strengthensthis view by showing the robust effect of the spontaneous activityof this region in predicting individual participants’ variation insemantic processing behavior.

Regions for which connectivity with LMTG predictedsemantic processing behaviorAmong the brain regions showing resting-state synchronizationwith the LMTG, a subset of these regions further predicted vari-ation in semantic processing behavior, although their contribu-tion beyond the regional ALFF effect of the LMTG was notsignificant. Among these regions, the LIFG and left ATL haveconsistently been observed to have white matter fiber pathways

Table 2. Clusters for which functional connectivity strength with LMTG can predict semantic performance

Brain regions

Regression coefficient Correlation coefficient

Peak MNI coordinates Peak MNI coordinates

BA x y z r ( p) (peak) Volumes (mm 3) BA x y z r ( p) (peak) Volumes (mm 3)

LIFG 11, 47 �30 54 �12 0.78 (�0.001) 2943 11, 47 �30 54 �12 0.62 (�0.001) 1593Left MTG 38, 37, 21, 20 �54 �54 3 0.71 (�0.001) 16038 20, 21 �54 �39 �6 0.46 (�0.01) 1593Right MTG 38, 37, 21, 20 51 18 �21 0.67 (�0.001) 11475 38, 21 45 3 �33 0.57 (�0.001) 2646Left MTL 20, 37 �30 �9 �18 0.64 (�0.001) 2565 20 �28 �20 �11 0.44 (�0.01) 324a

PCG 23 �3 �42 30 0.59 (�0.001) 1593 23 1 �41 30 0.40 (�0.05) 270a

DMPFC 8, 9, 10 15 63 21 0.65 (�0.001) 9936 9 6 51 45 0.53 (�0.01) 1647VMPFC 11 3 48 �15 0.58 (�0.001) 1755 11 18 60 �15 0.50 (�0.01) 1296

For the regression coefficients (�) results, a � map was generated for each participant by calculating the regression efficient between the seed time series and other voxels; then one sample t test on these individual � maps to see whetherthey were significantly different from zero (t � 6.43, corrected p � 0.01, Bonferroni correction); then for the significant regions, correlation analyses between participants’ semantic performance and � values were conducted whileregressing out the peripheral response components reflected in the performance of the control tasks. The correlation coefficient (r) results were obtained using the same procedure except that correlation efficient between the seed time seriesand other voxels were considered and that Fisher z-score transformation was performed on these r values.aThe peak voxel of these clusters reached the threshold of p � 0.05, but they did not survive multiple comparison correction cluster size �38 voxels (1026 mm 3).

Table 3. The results of ROI analysis for the number judgment task

ROI regions Correlation r ( p)

Regional activityLMTG 0.04 (0.82)

functional connectivityLIFG 0.03 (0.87)LMTG (including ATL) �0.03 (0.37)Right MTG (including ATL) �0.07 (0.69)LMTL �0.18 (0.30)PCG �0.08 (0.65)DMPFC 0.02 (0.91)VMPFC �0.11 (0.54)


(i.e., anatomical connectivity) projecting to the LMTG in diffu-sion tensor imaging (DTI) and anatomical dissection studies(Catani et al., 2005; Catani and Mesulam, 2008; Papagno et al.,2011; Turken and Dronkers, 2011). Importantly, these fiber bun-dles seem to be specifically engaged in semantic and not phono-logical tasks (Saur et al., 2008, 2010). Furthermore, LIFG,especially the specific region implicated in the current study (BA47), is activated by a variety of semantic tasks (Fiez, 1997; Pol-drack et al., 1999; Friederici et al., 2003). The involvement of theATL in semantic processing is most strongly suggested by seman-tic dementia patients, who suffer from atrophy of the ATL ac-companied by progressive semantic impairment (for review, seePatterson et al., 2007).

Using DTI, Gong et al. (2009) reported that white matter fiberbundles exist between the LMTG and PCG. PCG has been shownto be associated with various cognitive functions, including epi-sodic retrieval (Wagner et al., 2005), self-monitoring (Vogt andLaureys, 2005), self-reflection (Johnson et al., 2002), and aware-ness (Vogt and Laureys, 2005). We speculate that semantic pro-cessing might induce episodic encoding and thus can lead to PCGengagement.

The anatomical connections between the other three regions(MTL, VMPFC, and DMPFC) and LMTG are far from clear.Neurophysiological, neuroimaging, and neuropsychologicalstudies provide converging evidence that the MTL is critical forsemantic processing, especially the acquisition and retrieval ofsemantic memory (for review, see Squire et al., 2004). VMPFChas been strongly linked to social cognition. Specifically, it hasbeen ascribed a crucial role for the encoding of enduring socialdispositions and interpersonal knowledge through the integra-tion of social information over long stretches of time in multiplecircumstances (Van Overwalle, 2009). Such processes may beshared by semantic memory, which also implicates the abstrac-tion of consistent information from disparate instances of anobject or event and the decontextualization of individual mem-ories (Takashima et al., 2006). The role of DMPFC is less clear,

but, as speculated by Binder et al. (2009), it might be involved in“self-guided retrieval” aspects of semantic processing. Lesions tothis region have been found to induce symptoms of very littlespontaneous speech or action despite the preserved ability tospeak or act when prompted (Nagaratnam et al., 2004).

Finally, the resting-state semantic network observed herelargely overlaps with the default mode network (Shulman et al.,1997; Raichle et al., 2001). It has been proposed that during itsconscious resting state, this network reflects ongoing semanticprocessing, such as semantic knowledge retrieval and manipula-tion of represented knowledge for problem solving (Binder et al.,2009). Other researchers have linked the network’s activity tointernally generated thought processes related to self-reflection(Gusnard et al., 2001) or mind wandering (Mason et al., 2007)—processes that implicate a critical role for semantics. Our resultsfurther reinforce the notion that the function of the default modenetwork involves semantic processing.

Potential mechanisms for the rs-fMRI– behavior correlationThe neuronal origin of rs-fMRI activity has been supported bymany types of findings, including its correlation with electro-physiological recordings of neuronal firing (Nir et al., 2008),structural connectivity, slow cortical potentials, and the band-limited power of fast electrical activity (for review, see Zhang andRaichle, 2010). The correlation between rs-fMRI and behavioralefficiency reported here might be driven by either genetic and/orenvironmental variables. The influence of genetic factors on spe-cific rs-fMRI patterns has been demonstrated (Glahn et al., 2010),and it is not implausible to assume that variation in genetic dis-position may optimize brain regions/networks to acquire andprocess semantic knowledge most efficiently. On the other hand,training has also been shown to affect rs-fMRI patterns (Lewisa etal., 2009), and it is possible that the richness of semantic experi-ence shapes the rs-fMRI patterns in relevant regions. Future stud-ies should consider the relative contribution of these variables indetermining rs-fMRI patterns.

Figure 4. Default mode network and its comparison with the observed semantic network. A, Default mode network results. The bilateral inferior parietal lobe, medial prefrontal cortex, superiorfrontal cortex, medial temporal lobe, and lateral temporal cortex had strong functional connectivity with PCG, and these regions constitute the default mode network. B, Overlay of Figure 3 and A,showing that the semantic network lies within the default mode network.


In conclusion, we observed a network, with the LMTG as itscore component, in which the resting-state activity predictshealthy individuals’ conceptual processing efficiency. This find-ing has multifaceted theoretical implications: (1) it further con-firms the roles of these regions, especially the central role ofLMTG, in object conceptual processing; (2) it reveals a potentialnew mechanism for the representation and “maintenance” ofsemantic memory (i.e., through resting-state fluctuations); and(3) the extensive overlap between the resting state of the semanticnetwork and the default mode network reinforces previousclaims that the function of the default mode network may involvesemantic processing. Finally, our findings have important clinicalimplications, providing potential biomarkers for the early detec-tion of conceptual knowledge deterioration.

ReferencesBates E, Wilson SM, Saygin AP, Dick F, Sereno MI, Knight RT, Dronkers NF

(2003) Voxel-based lesion-symptom mapping. Nat Neurosci 6:448–450.Binder JR, Desai RH, Graves WW, Conant LL (2009) Where is the semantic

system? A critical review and meta-analysis of 120 functional neuroimag-ing studies. Cereb Cortex 19:2767–2796.

Biswal B, Yetkin FZ, Haughton VM, Hyde JS (1995) Functional connectiv-ity in the motor cortex of resting human brain using echo-planar MRI.Magn Reson Med 34:537–541.

Catani M, Mesulam M (2008) The arcuate fasciculus and the disconnectiontheme in language and aphasia: history and current state. Cortex44:953–961.

Catani M, Jones DK, Ffytche DH (2005) Perisylvian language networks ofthe human brain. Ann Neurol 57:8 –16.

Damasio H, Grabowski TJ, Tranel D, Hichwa RD, Damasio AR (1996) Aneural basis for lexical retrieval. Nature 380:499 –505.

Dehaene S, Piazza M, Pinel P, Cohen L (2003) Three parietal circuits fornumber processing. Cogn Neuropsychol 20:487–506.

Di Martino A, Shehzad Z, Kelly C, Roy AK, Gee DG, Uddin LQ, Gotimer K,Klein DF, Castellanos FX, Milham MP (2009) Relationship betweencingulo-insular functional connectivity and autistic traits in neurotypicaladults. Am J Psychiatry 166:891– 899.

Dronkers NF, Wilkins DP, Van Valin RD Jr, Redfern BB, Jaeger JJ (2004)Lesion analysis of the brain areas involved in language comprehension.Cognition 92:145–177.

Fiez JA (1997) Phonology, semantics, and the role of the left inferior pre-frontal cortex. Hum Brain Mapp 5:79 – 83.

Forster KI, Forster JC (2003) DMDX: a windows display program with mil-lisecond accuracy. Behav Res Methods Instrum Comput 35:116 –124.

Fox MD, Snyder AZ, Vincent JL, Corbetta M, Van Essen DC, Raichle ME(2005) The human brain is intrinsically organized into dynamic, anticor-related functional networks. Proc Natl Acad Sci U S A 102:9673–9678.

Friederici AD, Ruschemeyer SA, Hahne A, Fiebach CJ (2003) The role of leftinferior frontal and superior temporal cortex in sentence comprehension:localizing syntactic and semantic processes. Cereb Cortex 13:170 –177.

Glahn DC, Winkler AM, Kochunov P, Almasy L, Duggirala R, Carless MA,Curran JC, Olvera RL, Laird AR, Smith SM, Beckmann CF, Fox PT,Blangero J (2010) Genetic control over the resting brain. Proc Natl AcadSci U S A 107:1223–1228.

Gong GL, He Y, Concha L, Lebel C, Gross DW, Evans AC, Beaulieu C (2009)Mapping anatomical connectivity patterns of human cerebral cortex us-ing in vivo diffusion tensor imaging tractography. Cereb Cortex19:524 –536.

Gusnard D, Akbudak E, Shulman G, Raichle ME (2001) Role of medialprefrontal cortex in a default mode of brain function. Neuroimage13:S414 –S414.

Hampson M, Tokoglu F, Sun ZD, Schafer RJ, Skudlarski P, Gore JC, Consta-ble RT (2006) Connectivity-behavior analysis reveals that functionalconnectivity between left BA39 and Broca’s area varies with reading abil-ity. Neuroimage 31:513–519.

Hickok G, Poeppel D (2004) Dorsal and ventral streams: a framework forunderstanding aspects of the functional anatomy of language. Cognition92:67–99.

Hoptman MJ, Zuo XN, Butler PD, Javitt DC, D’Angelo D, Mauro CJ, Milham

MP (2010) Amplitude of low-frequency oscillations in schizophrenia: aresting state fMRI study. Schizophr Res 117:13–20.

Howard D, Patterson KE (1992) The pyramids and palm trees test: a test ofsemantic access from words and pictures. Bury St. Edmunds, UK: ThamesValley Test Company.

Johnson SC, Baxter LC, Wilder LS, Pipe JG, Heiserman JE, Prigatano GP(2002) Neural correlates of self-reflection. Brain 125:1808 –1814.

Koyama MS, Kelly C, Shehzad Z, Penesetti D, Castellanos FX, Milham MP(2010) Reading networks at rest. Cereb Cortex 20:2549 –2559.

Lau EF, Phillips C, Poeppel D (2008) A cortical network for semantics: (de-)constructing the N400. Nat Rev Neurosci 9:920 –933.

Lewisa CM, Baldassarre A, Committeri G, Romani GL, Corbetta M (2009)Learning sculpts the spontaneous activity of the resting human brain.Proc Natl Acad Sci U S A 106:17558 –17563.

Logothetis NK, Pauls J, Augath M, Trinath T, Oeltermann A (2001) Neuro-physiological investigation of the basis of the fMRI signal. Nature412:150 –157.

Maguire EA, Frith CD (2004) The brain network associated with acquiringsemantic knowledge. Neuroimage 22:171–178.

Martin A, Wiggs CL, Ungerleider LG, Haxby JV (1996) Neural correlates ofcategory-specific knowledge. Nature 379:649 – 652.

Mason MF, Norton MI, Van Horn JD, Wegner DM, Grafton ST, Macrae CN(2007) Wandering minds: the default network and stimulus-independentthought. Science 315:393–395.

Mennes M, Zuo XN, Kelly C, Di Martino A, Zang YF, Biswal B, CastellanosFX, Milham MP (2011) Linking inter-individual differences in neuralactivation and behavior to intrinsic brain dynamics. Neuroimage54:2950 –2959.

Mummery CJ, Patterson K, Price CJ, Ashburner J, Frackowiak RSJ, Hodges JR(2000) A voxel-based morphometry study of semantic dementia: rela-tionship between temporal lobe atrophy and semantic memory. AnnNeurol 47:36 – 45.

Nagaratnam N, Nagaratnam K, Ng K, Diu P (2004) Akinetic mutism fol-lowing stroke. J Clin Neurosci 11:25–30.

Nir Y, Mukamel R, et al (2008) Interhemispheric correlations of slow spon-taneous neuronal fluctuations revealed in human sensory cortex. NatNeurosci 11:1100 –1108.

Papagno C, Gallucci M, Casarotti A, Castellano A, Falini A, Fava E, GiussaniC, Carrabba G, Bello L, Caramazza A (2011) Connectivity constraintson cortical reorganizationof neural circuits involved in object naming.Neuroimage 55:1306 –1313.

Patterson K, Nestor PJ, Rogers TT (2007) Where do you know what youknow? The representation of semantic knowledge in the human brain.Nat Rev Neurosci 8:976 –987.

Poldrack RA, Wagner AD, Prull MW, Desmond JE, Glover GH, Gabrieli JDE(1999) Functional specialization for semantic and phonological process-ing in the left inferior prefrontal cortex. Neuroimage 10:15–35.

Raichle ME (2006) The brain’s dark energy. Science 314:1249 –1250.Raichle ME, MacLeod AM, Snyder AZ, Powers WJ, Gusnard DA, Shulman

GL (2001) A default mode of brain function. Proc Natl Acad Sci U S A98:676 – 682.

Rudrauf D, Mehta S, Bruss J, Tranel D, Damasio H, Grabowski TJ (2008)Thresholding lesion overlap difference maps: application to category-related naming and recognition deficits. Neuroimage 41:970 –984.

Saur D, Kreher BW, Schnell S, Kummerer D, Kellmeyer P, Vry MS, UmarovaR, Musso M, Glauche V, Abel S, Huber W, Rijntjes M, Hennig J, Weiller C(2008) Ventral and dorsal pathways for language. Proc Natl Acad SciU S A 105:18035–18040.

Saur D, Schelter B, Schnell S, Kratochvil D, Kupper H, Kellmeyer P, Kum-merer D, Kloppel S, Glauche V, Lange R, Mader W, Feess D, Timmer J,Weiller C (2010) Combining functional and anatomical connectivityreveals brain networks for auditory language comprehension. Neuroim-age 49:3187–3197.

Schwartz MF, Kimberg DY, Walker GM, Faseyitan O, Brecher A, Dell GS,Coslett HB (2009) Anterior temporal involvement in semantic word re-trieval: voxel-based lesion-symptom mapping evidence from aphasia.Brain 132:3411–3427.

Shulman GL, Fiez JA, Corbetta M, Buckner RL, Miezin FM, Raichle ME,Petersen SE (1997) Common blood flow changes across visual tasks. 2.Decreases in cerebral cortex. J Cognitive Neurosci 9:648 – 663.


Simmons WK, Reddish M, Bellgowan PSF, Martin A (2010) The selectivityand functional connectivity of the anterior temporal lobes. Cereb Cortex20:813– 825.

Squire LR, Stark CE, Clark RE (2004) The medial temporal lobe. Annu RevNeurosci 27:279 –306.

Stoeckel MC, Weder B, Binkofski F, Buccino G, Shah NJ, Seitz RJ (2003) Afronto-parietal circuit for tactile object discrimination: an event-relatedfMRI study. Neuroimage 19:1103–1114.

Takashima A, Petersson KM, Rutters F, Tendolkar I, Jensen O, Zwarts MJ,McNaughton BL, Fernandez G (2006) Declarative memory consolida-tion in humans: a prospective functional magnetic resonance imagingstudy. Proc Natl Acad Sci U S A 103:756 –761.

Townsend JT, Ashby FG (1983) Stochastic modelling of elementary psycho-logical processes. New York: Cambridge UP.

Tulving E (1972) Organization of memory (Tulving E, Donaldson W, eds).London: Academic.

Turken AU, Dronkers NF (2011) The neural architecture of the languagecomprehension network: converging evidence from lesion and connec-tivity analyses. Front Syst Neurosci 5:1.

Tzourio-Mazoyer N, Landeau B, Papathanassiou D, Crivello F, Etard O, Del-croix N, Mazoyer B, Joliot M (2002) Automated anatomical labeling of

activations in SPM using a macroscopic anatomical parcellation of theMNI MRI single-subject brain. Neuroimage 15:273–289.

Vandenberghe R, Price C, Wise R, Josephs O, Frackowiak RS (1996) Func-tional anatomy of a common semantic system for words and pictures.Nature 383:254 –256.

Van Overwalle F (2009) Social cognition and the brain: a meta-analysis.Hum Brain Mapp 30:829 – 858.

Vogt BA, Laureys S (2005) Posterior cingulate, precuneal and retrosplenialcortices: cytology and components of the neural network correlates ofconsciousness. Prog Brain Res 150:205–217.

Wagner AD, Shannon BJ, Kahn I, Buckner RL (2005) Parietal lobe contri-butions to episodic memory retrieval. Trends Cogn Sci 9:445– 453.

Willms JL, Shapiro KA, Peelen MV, Pajtas PE, Costa A, Moo LR, CaramazzaA (2011) Language-invariant verb processing regions in Spanish–Eng-lish bilinguals. Neuroimage 57:251–261.

Yan C-G, Zang Y-F (2010) DPARSF: A MATLAB toolbox for “pipeline”data analysis of resting-state fMRI. Front Syst Neurosci 4:13.

Zang YF, He Y, Zhu CZ, Cao QJ, Sui MQ, Liang M, Tian LX, Jiang TZ, WangYF (2007) Altered baseline brain activity in children with ADHD re-vealed by resting-state functional MRI. Brain Dev 29:83–91.

Zhang D, Raichle ME (2010) Disease and the brain’s dark energy. Nat RevNeurol 6:15–28.


behavioral/systems/cognitive ... · a cued articulation task con-sisting of 20 trials was the...

Documents